The classic recursive least-square(RLS) fine alignment method, which uses the horizontal velocity errors as measurements, needs longer time to identify the azimuth error and gyro drift errors. Besides, its convergent speed is affected by the initial value of recursive calculation. An improved method is put forward which directly uses the horizontal specific forces in one second as measurements, and neglects high-order misalignment errors in converting the cubic curve into linear model. On the basis of the simplified model, a low computational complexity RLS algorithm without initial condition is derived to speed up the convergence. The simulation and practicality experiments based on stationary base show that the improved method can identify the misalignment errors and the north gyro drift errors in 120 s, so it greatly shortens the time of initial alignment and gyro drift measurement without affecting alignment accuracy.%传统最小二乘精对准方法采用水平速度误差作为量测值,对方位失准角和陀螺漂移的估计时间较长,且收敛速度受递推计算初值影响.改进方法直接采用单位时间的水平比力作为量测值,同时忽略失准角误差中的高次项,将三次曲线转化为线性模型进行估计.在简化模型的基础上,推导了计算量很小的无初值最小二乘递推算法,进一步加快了对准收敛速度.静基座条件下的仿真和实验验证结果表明,改进方法估计的失准角误差和北向陀螺漂移均可在120 s内收敛,在不影响对准精度的前提下缩短了对准和测漂时问,
展开▼
